Role of Artificial Intelligence in the Diagnosis and Management of Pulmonary Embolism: A Comprehensive Review
Ahmad Moayad Naser, Rhea Vyas, Ahmed Ashraf Morgan, Abdul M. Kalaiger, Amrin Kharawala, Sanjana Nagraj, Raksheeth Agarwal, Maisha Maliha, Shaunak Mangeshkar, Nikita Singh, Vikyath Satish, Sheetal Vasundara Mathai, Leonidas Palaiodimos, Robert Faillace
Abstract
Pulmonary embolism (PE) remains a critical condition with significant mortality and morbidity, necessitating timely detection and intervention to improve patient outcomes. This review examines the evolving role of artificial intelligence (AI) in PE management. Two primary AI-driven models that are currently being explored are deep convolutional neural networks (DCNNs) for enhanced image-based detection and natural language processing (NLP) for improved risk stratification using electronic health records. A major advancement in this field was the FDA approval of the Aidoc© AI model, which has demonstrated high specificity and negative predictive value in PE diagnosis from imaging scans. Additionally, AI is being explored for optimizing anticoagulation strategies and predicting PE recurrence risk. While further large-scale studies are needed to fully establish AI's role in clinical practice, its integration holds significant potential to enhance diagnostic accuracy and overall patient management.